Database column types
Simple variables and record structures can be defined from database columns types.
Variables defined with the LIKE
keyword get the same data type as the table
column of a database
schema.
SCHEMA stores
DEFINE cust_name LIKE customer.cust_name
MAIN
DEFINE rec_cust RECORD LIKE customer.*
...
END MAIN
A SCHEMA
statement must define the database name identifying the database schema
files to be used.
DEFINE rec_city RECORD LIKE base:city.*
DEFINE rec_country RECORD LIKE base:country.*
DEFINE rec_customer RECORD LIKE orders:customer.*
DEFINE rec_item RECORD LIKE stock:item.*
At runtime, a program typically connects to a single database source. Using multiple database schemas is a programming feature.
The database schema files must exist and must be located in one of the directories specified in the FGLDBPATH environment variable.
For CHAR/VARCHAR
types, the size in the .sch
file, the size in
the .sch file is expressed in character units, and is then interpreted by the
compiler as a number of bytes or characters, depending on the FGL_LENGTH_SEMANTICS
environment variable as when using directly CHAR(N)
or VARCHAR(N)
.
For more details, see Extracting database schemas.
When using database views, the column cannot be based on an aggregate
function like SUM()
.
If LIKE
references a SERIAL
column,
the variable will be defined with the INTEGER
data
type. If LIKE
references an INT8
, SERIAL8
or BIGSERIAL
column,
the variable will be defined with the BIGINT
data
type.
The table qualifier must specify owner if table.column is not a unique column identifier within its database, or if the database is ANSI-compliant and any user of your application is not the owner of table.